ms2: A Molecular Simulation Tool for Thermodynamic Properties

More documents

Recommendations

Info

$Nanomechanical strength and fracture mechanisms of hierarchical ...$

a vapor density of 0.443 mol/l. Using Towhee, the calculation yielded a vapor pressure of 1.81 MPa as well as densities of 13.62 mol/l and 0.64 mol/l, respectively. Here, the standard deviations of the results were significantly higher. These simulation results showed a drift, e.g. for the saturated liquid density by an average of 1.50 mol/l over 1000 steps. The results for the Errington Code and MCCCS Towhee indicate that the simulations with these programs had not reached the phase equilibrium in 96 h. This effect is much more pronounced forMCCCS Towhee, having executed significantly less loops than the Errington code. Molecular dynamics. The performance of the MD part of ms2 was evaluated against the simulation program Gromacs V4.0.3 [69], which is designed for simulations of biological systems. The comparison was restricted to a simple NVT simulation of the equimolar liquid mixture of ethanol and methanol at T = 298 K and p = 0.1 MPa. The runs were executed sequentially on a ”Harpertown”/Infiniband cluster with an equal number of time steps and the same interaction cut-off radius of 21 ˚A. Both programs,ms2 andGromacs, were compiled with the Intel ifortran compiler. Gromacs was faster by almost a factor of two than ms2 and also the scaling behavior for the parallel version was superior. This shows that there is still an optimization potential in ms2, which will be exploited in the future. A comparison between Gromacs and ms2 for the VLE test case was not carried out, since Gromacs does not support VLE simulations. 6.6. Computational demand for the calculation of transport properties The computational demand for transport properties following the Green-Kubo formalism was evaluated. This investigation was performed for the test case equimolar liquid mixture of methanol and ethanol at a temperature T = 298 K andp = 0.1 MPa with an autocorrelation length of 13.82 ps. The time period between two autocorrelation functions was specified to be 197.4 fs. A total of 100 autocorrelation functions was explicitly written to file in order to check the results. The transport properties were evaluated every four autocorrelation functions. The reference case was a MD run performed with the same technical details, except for the calculation of the autocorrelation functions. Calculating transport properties increased the CPU time of a simulation step by about 78% compared to the reference case. Note that all autocorrelation functions were evaluated each time step, which is an extreme frequency that can be reduced without much loss of accuracy. 7. Features The feature programs described in the following are intended to facilitate the handling of ms2. They should allow for an easy start of molecular simulations with ms2 and give access to all output data generated byms2. 7.1. Simulation setup: ms2par The GUI-based feature tool ms2par allows for a convenient generation of input (*.par) files according to user specifications, cf. Figure 10. It assigns the molecular model, ensemble, thermodynamic state point, time step, number of equilibration steps etc. Furthermore, the tool proposes a cut-off radius based on the input data. After the user completes the specifications, the program generates the *.par-file in ASCII format, to be read by ms2. 24
Applyingms2par is simple and should be intuitive even for new users. ms2par is a java application, thus it runs on all operating systems. 7.2. Simulation analysis: ms2chart ms2chart is a GUI-based java applet for evaluating the simulation results from ms2, cf. Figure 11. This tool plots the evolution of properties calculated by ms2. The properties can be shown for different graph axes that are chosen from a drop down menu and the plot is shown directly in the GUI. For a better analysis, ms2chart allows various features: plotting block averages as well as simulation averages into the same plot, changing the design of the plot, individual labeling of the axes and adjusting the frame detail. All plots can be exported in png format. The analysis program ms2chart can be executed at any time of the simulation, i.e. also while the simulation is still running. There is no loss of data, if it is run on the fly. The tool ms2chart is easy to understand and can be handled intuitively. It allows for an easy first analysis of the ms2 simulation results. Users will employ ms2chart for a quick check of their simulation results, such as whether the equilibration time was appropriate, which is important if extensive series of simulation runs are performed. 7.3. Visualization tool: ms2molecules The program ms2molecules is a visualization tool for ms2. The program displays the molecular trajectories that are stored by ms2 as a series of configurations in the *.vim file. ms2molecules visualizes molecular sites by colored spheres. The colors are user defined. The size of a sphere is by default proportional to the LJ size parameterσ, but can be changed manually in the first lines of the *.vim file. This feature facilitates monitoring a component of interest by reducing the size of other components, e.g. solvents. Other features like zooming into or out of the simulation volume as well as rotating the simulation volume also facilitate the analysis of simulations. The visualization can be exported via snapshots in the jpg format. The program is based on OpenGL and written in C. The handling of the program is simple and console based. Requirements for the feature tool are OpenGL in a Windows or Linux environment. Figure 12 shows a snapshot of a ternary mixture, taken with the program ms2molecules, which is convenient to gain insight into the trajectory of the system or the state of the system, respectively. 8. Conclusions The molecular simulation program ms2 was designed for the calculation of thermodynamic properties of bulk fluids. Special care was given to a minimization of the response time. The capabilities of ms2 are broad, rang- ing from basic static thermodynamic properties, like thermal and caloric data, over vapor-liquid equilibria to transport properties, like diffusion coefficient, viscosity and thermal conductivity. These data are accessible for pure substances and mixtures. The accuracy of the simulation data generated by ms2 is high, while consum- ing a reasonable computational effort. Molecular models of more than 100 pure fluids are supplied with ms2 that accurately describe their thermodynamic properties. Despite the fact that ms2 is a sophisticated Fortran90 program, new developers benefit from its modular structure and object-orientation. The application of ms2 is 25
Page 1 and 2: ms2: A Molecular Simulation Tool fo
Page 3 and 4: 2. Introduction Due to the advances
Page 5 and 6: 3. Simulation programms2 The simula
Page 7 and 8: sense that the specified chemical p
Page 9 and 10: 3.3.3. Chemical potential The chemi
Page 11 and 12: 3.3.6. Transport properties In ms2,
Page 13 and 14: −0.5 < x, y, z < 0.5. The advanta
Page 15 and 16: Q = 2aq. The electrostatic interact
Page 17 and 18: Programming (OOP). It should be men
Page 19 and 20: motion, kinetic energy calculation,
Page 21 and 22: • PGI pgf95 4 with ”-r8 -fastss
Page 23: memory scales linearly with the num
Page 27 and 28: Table 1: Computational demand for t
Page 29 and 30: Table 3: NEC SX8R ”ftrace” prof
Page 31 and 32: i r ij ij Figure 2: Schematic of t
Page 33 and 34: Figure 4: UML class diagram of ms2,
Page 35 and 36: Figure 6: Runtime for the equimolar
Page 37 and 38: (a) (b) Figure 8: ms2 runtime of th
Page 39 and 40: Figure 10: Snapshot of ms2par gener
Page 41 and 42: Figure 12: Snapshot of a ternary mi
Page 43 and 44: [24] Eckl, B., Vrabec, J., and Hass
Page 45 and 46: Supplementary Material to ms2: A Mo
Page 47 and 48: The partial derivative of the resid
Page 49 and 50: 2. Details ofms2 This section provi
Page 51 and 52: where u µRF i is its contribution
Page 53 and 54: 3. Modular structure - example The
Page 55 and 56: Table 2 continued Parameter Option
Page 57 and 58: An example for a *.par file is give
Page 59 and 60: Table 4: Parameters and options spe
Page 61: References [1] Allen, M. and Tildes

ms2: A Molecular Simulation Tool for Thermodynamic Properties

Create successful ePaper yourself

Delete template?

Save as template?